Session

DescriptionThis session is about closing integrity faster by pairing physics-grounded models with learning-based methods that improve scalability and decision turnaround in real design flows. The papers span joint EM–IR modeling, robust simulation/solver techniques, and physics-informed learning that supports earlier screening and design-space exploration without disconnecting from electrical behavior. On the thermal side, the session highlights modular chiplet-oriented thermal modeling and cooling-aware prediction to surface hotspots and constraints earlier in the flow. Complementing these are constraint-aware P/G optimization and energy modeling for agile accelerator design. Together, these contributions emphasize practical, actionable integrity insights earlier, helping teams converge with fewer late surprises.